Segmentasi Exudate pada Citra Fundus Menggunakan Mathematical Morphology dan Kombinasi Renyi Entropy Thresholding dengan Cuckoo Search Optimization Algorithm.

Qomariah, Dinial Utami Nurul (2018) Segmentasi Exudate pada Citra Fundus Menggunakan Mathematical Morphology dan Kombinasi Renyi Entropy Thresholding dengan Cuckoo Search Optimization Algorithm. Masters thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
5114201026-Buku.pdf - Published Version
Restricted to Repository staff only

Download (1MB) | Request a copy

Abstract

Diabetic retinopathy merupakan penyakit yang disebabkan oleh diabetes, yang menyebabkan abnormalitas dari pembuluh darah retina. Salah satunya exudates yang merupakan lapisan lemak dan protein yang pecah pada pembuluh darah yang abnormal dan bisa menyebabkan kebutaan bila berada pada area macula. Sebelum mendeteksi exudates, area optic disk pada retina perlu dideteksi dan dihilangkan, karena optic disk juga memiliki nilai intensitas yang hampir sama dengan exudates. Area seperti pembuluh darah dan haemorhage juga perlu dihilangkan karena area tersebut tidak berkaitan dengan proses segmentasi exudates. Penelitian ini mengusulkan sebuah metode untuk segmentasi exudates pada citra fundus menggunakan mathematical morphology dan kombinasi renyi entropy thresholding dengan cuckoo search optimization algorithm. Mathematical morphology untuk mendeteksi exudates, akan tetapi metode morfologi saja tidak cukup untuk mengurangi over segmentasi sehingga untuk mengurangi oversegmentasi dikembangkan metode renyi entropy thresholding yang mempertimbangkan intensitas dari gambar. Metode renyi entropi thresholding dikombinasikan dengan metode optimasi cuckoo search algorithm. Pengembangan metode renyi entropy thresholding dilakukan agar dapat menghasilkan nilai threshold lebih optimal dengan mengoptimalkan nilai parameter rho pada renyi entropy yang ditetapkan antara nilai 0-1 menjadi adaptif menggunakan pendekatan cucko search algorithm. Sehingga metode yang dihasilkan menjadi renyi entropy thresholding berdasarkan cucko search optimization algorithm. Pengujian dilakukan pada gambar diaretdb1. Dataset diolah berdasarkan metode yang diajukan dengan menghitung nilai sensitivity, specificity dan accuracy dengan nilai berturut-turut 92,26%, 99,77% dan 99,72%. ============================================================ Diabetic retinopathy is a disease caused by diabetes, which is caused by abnormalities of the blood vessels in the eyes. One of them are exudates which are fat that broken on the abnormal blood vessels and can lead to blindness. Before detecting exudates, optic disk area is detected and removed since it has similar intensity with exudates. Mathematical morphology is used to detect the area of exudates and remove the area of the optic disk, because the morphological process still results oversegmentation then thresholding is done to reduce over-segmentation. Area such as blood vessels and haemorhage need to be removed because the area is not related to the exudates segmentation process. This research proposes a method for segmentation of exudates on the fundus image using mathematical morphology and renyi entropy thresholding combination with cuckoo search optimization algorithm. Mathematical morphology is to detect exudates, but morphological methods are not sufficient to reduce over segmentation. The over-segmentation is reduced using renyi entropy thresholding method which counts value of the intensity of the image. Renyi entropy thresholding method combined with cuckoo search algorithm optimization method. The method of renyi entropy thresholding is employed in order to get more optimal threshold value by optimizing the value of rho parameter on renyi entropy which the value is between 0-1 to be adaptive using cuckoo search algorithm approach. So the method becomes renyi entropy thresholding based on cuckoo search optimization algorithm. The test is performed on diaretdb1 image. The dataset is processed by the proposed method by calculating the sensitivity, specificity and accuracy value with the results 92.26%, 99.77% and 99.72% respectively.

Item Type: Thesis (Masters)
Additional Information: RTIf Qom s
Uncontrolled Keywords: citra fundus, exudates, mathematical morphology, renyi entropi thresholding, cuckoo search, segmentasi.
Subjects: Q Science > QA Mathematics > QA274.7 Markov processes--Mathematical models.
Q Science > QA Mathematics > QA76.9 Computer algorithms. Virtual Reality. Computer simulation.
Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis
Divisions: Faculty of Information Technology > Informatics Engineering > (S2) Master Theses
Depositing User: Dinial Utami Nurul Qomariah
Date Deposited: 28 Feb 2018 08:45
Last Modified: 28 Feb 2018 08:45
URI: http://repository.its.ac.id/id/eprint/49655

Actions (login required)

View Item View Item